AI Hunts Tax Evaders: 354 Million Euros Recovered

Artificial Intelligence (AI) offers countless applications. AI methods are already being used to combat tax fraud - with such methods, additional tax revenues of 354 million euros were recorded last year, the Ministry of Finance reported. There is a dedicated department, the Predictive Analytics Competence Center (PACC), which helps increase additional revenues through audit measures. With risk models, there were 6.6 million audits in 2024. An expansion is imminent.

Dedicated Competence Center with 6.6 Million Audits

The PACC found success "in all areas of financial administration," according to the Ministry of Finance to the APA. This ranges from fraud in employee assessments to income, corporate, and sales tax fraud. Millions of potential compliance violations have also been examined. This involves, for example, the identification of unjustly claimed and received subsidies and benefits, as well as shell companies. Given the budget deficit, tax fraud is particularly painful.

Marterbauer Announces "Tax Fraud Combat Package"

"Just as fraudulent energies constantly change and evolve, fraud prevention must also employ modern methods," said Finance Minister Markus Marterbauer (SPÖ), who announces the use of even more AI in the future and the presentation of a "Tax Fraud Combat Package" again in the fall. It is about justice: "Those who cheat not only harm the state but all honest taxpayers," said Marterbauer.

According to PACC head Christian Weinzinger, advanced predictive analytics and machine learning methods of AI contribute to the department's success. They cover a relatively broad spectrum of various topics related to tax audits, customs control, and other areas. "These measures relate both to the past and to the detection of potential risks in real-time or in advance," said Weinzinger. This includes standardized audit procedures for accounting documents, cryptocurrencies, and international data exchange programs.

(APA/Red)

This article has been automatically translated, read the original article here.